Archive tools that never upload the archive

A container and a compressor are not the same thing

Two separate jobs get muddled together under the word "archive". Bundling is putting many files into one file, keeping their names and folders. Compression is making bytes smaller. TAR only bundles — a .tar is the size of everything in it added together. Gzip only compresses, and only one stream at a time, which is why .tar.gz exists: bundle first, then squeeze the bundle. ZIP does both at once, per file, and keeps a catalogue of what is inside.

That difference decides how the format behaves under you. Because a ZIP stores a catalogue at the end and compresses each file separately, it can list a ten-gigabyte archive instantly and unpack one file out of the middle without touching the rest — which is exactly what our ZIP reader does, decompressing nothing until you ask for something. A solid archive compresses everything as one continuous stream, so repetition across file boundaries is found and the result is smaller, sometimes much smaller for many similar small files such as source code or logs. The price is that there is no cheap way to reach one file in the middle: everything before it has to be decompressed first. TAR.GZ is solid by nature, 7z compresses in solid blocks, and a RAR is read through from the start here rather than from a catalogue.

Already-compressed content does not compress again. JPEG, PNG, MP3, MP4, PDF and every modern Office file have compression built in, and deflate spends time on them to achieve nothing — occasionally adding a few bytes for the trouble. This is why a zip of a photo folder comes out the same size as the folder and people assume the tool is broken. Create ZIP Archive recognises those formats and stores them as they are instead of pretending, which is why it finishes immediately on photographs and takes its time on text.

The metadata an archive carries is not the metadata your disk has. A ZIP stores modification times in the MS-DOS format, which has two-second resolution and cannot represent a date before 1980 or after 2099 — dates outside that range are clamped here rather than being allowed to fail the whole archive, which is a real bug we hit converting RARs that carried a zero date. A TAR stores Unix permissions, ownership and symbolic links. A browser can read none of those from your disk and cannot write any of them back, so Create TAR Archive writes a sensible default mode, and our extractors list permissions and links rather than restoring them. If an archive is a backup that has to come back exactly as it went in, it needs to be made and restored by tar on the machine itself.

ZIP encryption comes in two kinds and only one of them is worth anything. The original ZipCrypto scheme is broken to the point of being decorative — a known-plaintext attack recovers the contents in seconds on ordinary hardware, and most software still offers it by default. AES-256, added later, is real encryption and is not breakable. Neither is offered here: we will not ship the broken one, and the modern one has no browser-native implementation we can use under a permissive licence. Worth knowing either way: a ZIP never encrypts its file names, so the catalogue of an encrypted archive is readable by anyone. "Invoices/2024/Settlement-confidential.pdf" is visible without the password.

An archive is the one kind of file whose contents can attack the program opening it. A file inside can be named "../../.bashrc" to escape the folder it is unpacked into, and a few kilobytes can be crafted to expand into terabytes and take the tab down. Both are handled before you see anything: hostile names — parent-directory climbs, absolute paths, Windows drive letters, reserved device names, invisible characters used to disguise an extension — are clamped and every correction is listed, and expansion ratios are measured and refused rather than attempted. Symbolic links inside an archive are shown but never followed.

Where a browser is genuinely the wrong tool

Running in your browser is what makes these private, and it costs something. We would rather say so here than have you find out halfway through a job.

Very large archives. Everything is held in your device's memory, so the ceiling is the tab's memory budget rather than a plan limit. Listing a big ZIP is cheap because the catalogue is read on its own, but extracting is not: a multi-gigabyte archive will struggle on a phone and may fail on a laptop. 7-Zip, GNU tar and bsdtar stream from disk and do not care how big the file is.

Making a RAR, or a 7z. Neither is possible here, for different reasons. RAR compression is proprietary and the licence on the only code that creates it forbids using it to build another archiver, so no browser tool anywhere can honestly offer it — that is what the ZIP to RAR page exists to explain. For 7z, reading was possible through an independently written implementation and writing has no such path yet. If you need either, WinRAR makes the first and 7-Zip — free, and better at this than any web page — makes the second.

Anything encrypted. We cannot create a password-protected archive and we cannot open one. 7-Zip with AES-256 is the free answer for making one, and it has a further advantage over any web page: your password is typed into a program on your own machine rather than into a text box on the internet.

Faithful backups. Permissions, ownership, symbolic and hard links, extended attributes, sparse files and exact timestamps are what a backup is for, and a browser has access to none of them. GNU tar and bsdtar are free, installed already on macOS and Linux, and are the right tool.

Automation, and squeezing hard. These tools run when a person clicks. Archiving a directory every night belongs in a command line — tar, zip, 7z and zstd are all free. And if size is what matters, deflate is a 1990s algorithm: zstd at a high level or xz will beat a ZIP of the same files substantially, at the cost of a format your recipient may not be able to open.

Choosing between the tools that sound alike

ZIP Extractor vs RAR Extractor vs 7Z Extractor vs TAR Extractor. In practice, whichever you land on. Each one identifies the file by its first few bytes rather than its extension, so all four open all of the formats and tell you what the file really was — which matters, because a RAR renamed to .zip to get past a mail filter is common. What genuinely differs is the format, not the page: a ZIP lists instantly from its catalogue, while a RAR has to be read through once before it can tell you what is inside.

Create ZIP Archive vs Create TAR Archive. ZIP if a person is going to open it — it works on every computer with nothing installed. TAR.GZ if a machine or a Linux user is, or if you have many small similar files, where compressing the bundle as one stream is measurably smaller than compressing each file on its own. Plain TAR, with no compression at all, is only worth choosing when the contents are already compressed or something downstream will do the squeezing.

RAR to ZIP vs ZIP to RAR. One is a conversion and the other is an explanation. RAR to ZIP genuinely reads your RAR and writes a ZIP, with the files copied across byte for byte rather than re-encoded. ZIP to RAR is a page that tells you why nobody can do that, because the alternative was leaving the search term to sites that will either upload your files or hand you a renamed ZIP.

Create ZIP Archive vs zipping to save space. They are not the same goal. Bundling many files into one for sending works on anything. Making a large file smaller usually does not, because the large files people want to shrink — video, photographs, PDFs — are already compressed. For those, the image, video and PDF compressors on this site re-encode the content, which is the only thing that actually makes them smaller.